A. Malekpour, B.W. Karney, and B.J. Adams (Canada)
Genetic algorithm, simulation, flow control, pumping.
An improved on-off strategy for pump control is proposed for Iran’s Yazd water conveyance system. To evaluate the proposed logic, an unsteady hydraulic model is developed using quasi-steady theory with flow control introduced at the boundaries. The hydraulic behavior of the system is evaluated over a 24 hour period. Although the control logic is effective, the number of on-off cycles of the pumps in low operating flow appears to be high for such a long pipeline. This cycling problem can be improved by increasing the volume of the reservoirs fed by each pump station. A GA optimization finds the optimum expansion for each reservoir; since the hydraulic constraints could not be defined by analytical functions, the flow model is embedded into the optimization routines to record the number of on-off cycles of each pump and to check whether any constraints are violated. By limiting the on off cycles of each pump to just one per day, the optimal solution is quickly achieved by the GA during the first 30 generations. Comparing to an enumeration technique, the GA solution is captured with considerably reduced computational expense.
Important Links:
Go Back
